MFF baseline collection
DATA available at http://www.mathworks.com/matlabcentral/fileexchange/50938-a-benchmark-case-for-statistical-process-monitoring-cranfield-multiphase-flow-facility
CVA.py: CVA is achieved by python according to the source code (Matlab). However, There are differents in performances between the code replication and source code, which may be due to differences in inverse and SVD decomposition between Matlab and Python
VICVA: CVA trained by VI from the perspective of probabilistic modeling. Because of the limitation of our computers, downsampling is performaned to reduce the computation burden. Detailed introductions about model training and process monitoring are presented in ./tutorial/vi_cva.md